A study conducted by the scientific journal Nature Medicine has found that AI chatbots miss over half of medical diagnoses. The study, which asked 1,298 participants in the UK to use a large language model such as ChatGPT or Meta's Llama 3, correctly identified medical conditions in fewer than 34.5% of cases. However, the problem was discovered when participants tried to replicate the results using different questions. Initial messages often contained only partial information, suggesting that conversing with chatbots did not improve the probability of receiving a correct medical diagnosis.